I’ve thought about this question a few times but haven’t found a good solution yet. The problem needs more attention.
AI and computer recognition of images is getting better. I’ll go ahead and say that the probability of reaching the level where computers are as good at humans at any of kind of CAPTCHA/verify-I’m-not-a-bot-test within the next 20 years is very, very high. CAPTCHAs and other newer anti-bot measures cannot continue to work.
We don’t want our comment sections, discussion forums and social media full of spamming bots, so how do we keep them from doing so? I have three proposal types for the online world.
Make sending cost
Generally speaking, spam does not work well. The click-rate is very, very low but it pays because sending mass amounts of it is very cheap. In the analog world, we also see spam in our mail boxes (curiously, this is allowed but online spam is not!), but not quite as much as online. The cost of sending spam in the analog world keeps the amount down (printing and postage).
Participation in the online world is generally free after one has paid one’s internet bill. Generally, stuff that isn’t free on the internet is not used much. Free is the new normal. Free also means the barrier for participation is very low which is good for poor people.
The idea is that we add a cost to writing comments, but keep it very low. Since spam only works when sending large amounts (e.g. 10,000,000 emails per year), and normal human usage does not require sending comparably large amounts (<1,000 emails in most cases), we could add a cost to this which is prohibitively large for bots, but (almost) negligible for humans. E.g. .01 USD per email sent. Thus, human usage would cost <10 USD per year, but botting would cost 100,000 USD.
Who gets the money? One could make it decentralized, so that the blog/newspaper/media owner gets the money. In that way, discussing on a service also supports them, altho very little.
This could maybe work for email spam, but for highly read comment sections (e.g. on major newspapers or official forums for large computer games), the rather small price to pay for writing 1000 (or even 10) posts would not be a deterrent. Hence the pay-for-use proposal fails to deal with some situations.
Verified users by initial cost
Another idea based on payment is that one can set up a service where users can pay a small fee (e.g. 10 USD) to be registered. The account from this service can then be used to log into the comment section (forum etc.) of other sites and comment for free. The payment can be with cryptocurrency as before so anonymity can be preserved. It is also possible to create multiple outward profiles from one registered account so that a user cannot be tracked from site to site.
If an account has been found to send spam, it can then be disabled and the payment has been wasted. The payment will not have to be large, but it needs to be sufficient to run the service. Perhaps one can outsource the spammer-or-not decision-making to a subset of users who wish to work for free (many services rely upon a subset of users to provide their time, e.g. OKCupid).
The proposal has the same problem as the one above in that it requires payment to participate.
Verified users without payment
A third proposal is to set up a service where one can make a profile for free, but that requires one to somehow prove that one is a real person. This could be done with confidential information about a person e.g. passport + access to a database of this. This would probably require cooperation with officials in each country. Probably they will keep the information about who is who if they can, so it is difficult to see how privacy could be preserved with proposals of this type.
As before, accounts will still need to be deactivated if they are found to be spamming. If the government is involved, they will surely push for other grounds for deactivation: intellectual monopoly infringement, sex work related stuff, foul language, national security matters and so on. This makes it the least preferred solution type to me.
Generally, the goal is to preserve privacy, no cost of participation and being spam-free. How can it be done? Is online discussion doomed to be overspammed?